Subscribe Bookmark


Jun 23, 2011

Digging into my diet and fitness data

If you’re a regular reader of the JMP Blog, then you already know that those of us who work for JMP have taken a page from the Hair Club for Men. From our hobbies to internal activities, the people who work at JMP are also JMP users! I seriously considered using that classic line from Hair Club for Men commercials while preparing my poster titled “Analysis of Personal Diet and Fitness Data With JMPfor the upcoming Discovery Summit 2014 conference in Cary, NC.


Over my next few blog posts, I will be sharing some of what I have learned while preparing my Discovery Summit poster along with a few reflections on what I have learned during the first year of what I like to call “the PhD of me.”


My interest in self-tracking grew from a long struggle with my own weight and emotional eating habits. As an overweight middle school student, I discovered that tracking my meals and strength training workouts in a notebook helped me reach a healthy weight. Unfortunately, during stressful periods of my life, I often returned to food for stress relief. I gained 30 pounds during college and lost it in graduate school, only to gain 60 pounds when my first pregnancy coincided with end of my degree program. I have experienced several large weight swings since that time, yet I reached my goal of being in a healthy weight range by the time I entered my second pregnancy in early 2011. Within the first 6 months after my son's birth, I lost all the weight I had gained, and tracking my diet and workouts was an integral part of the process. I have maintained my weight within a much smaller range since spring 2012. While maintaining, I find that continuing to track my eating habits and activity level helps me stay mindful and avoid the patterns that caused me trouble in the past.


Wouldn’t it have been better just to share a graph instead of writing the paragraph above? As you may already know, JMP 12 offers me new tools to do exactly that!


Weight Graph Grad School to Present 9-9-14


Over the years, I have used a variety of heart rate monitors, pedometers, pulse and blood pressure monitors, though having to take and collate manual notes on those measures has been a hurdle in getting more of my data into analysis-ready format. I have been thrilled with the evolution of activity monitors and smart phone apps capable of collecting data passively without the need for extensive note-taking. In fact, the rise of activity monitors has fueled a whole movement called the Quantified Self (QS) that includes people like me who track activities ranging from diet and fitness information to Internet use, sleep, stress levels or other measures.


It may sound a bit weird or obsessive to people who don’t track information about themselves, but many QS fans find daily data collection incredibly useful in identifying and optimizing their dietary habits and daily routines. QS data can even be useful in health related pursuits. Some have used it to successfully pinpoint mysterious food or environmental allergy triggers. This past spring, my dad sent me a link to Gary Wolf’s 2010 Ted talk, and his description of the QS movement sounded immediately familiar to me. I identify with the QS movement more than ever after nearly four years of using a BodyMedia® FIT armband activity tracker and its food-logging software.


Like many users of such devices, I depend on the daily dashboards, weekly and monthly reports provided by the monitor’s web and app-based software to see short-term trends. I never seriously considered getting my data out of the tracking software and into JMP until I had accumulated years’ worth of food logging and activity data. Unfortunately, the longest time frame I could specify when exporting my activity data or food log information was 28 days. After importing my activity data from multi-worksheet Excel files interactively once with just two years’ worth of exported data files, I concluded that I would need to automate this process through scripting.


Like it is for so many of our customers, JMP was the perfect tool to help me move beyond standard reports to truly exploring my data.  If I had realized how much I would learn from tackling this seemingly unrelated analysis project, I would have started it sooner! I mapped out the steps I needed to take to get my activity summary data in from each of the multiple worksheets and collected snippets of JSL code, including a very helpful loop example from a SESUG paper written by JMP Mac developer Michael Hecht. Soon, I was able to merge, clean and format my combined data table.


The scripting experience I gained from successfully tackling the Excel file import helped me with my next challenge: importing my food log files. While I hoped to use a PDF-to-Excel conversion program, I found the structure of the PDF tables in the BodyMedia® files was not regular enough to convert cleanly to Excel.


I converted my food log files to text instead and imported them into JMP. With advice from JMP developer Craige Hales, I parsed out the information I needed using the JSL-based regular expression engine in JMP. When I got stuck, I depended heavily on online scripting resources and helpful suggestions from resident JSL experts Melanie Drake, Rosemary Lucas and Audrey Ventura. Upon completing the project, I decided to submit an abstract for Discovery Summit covering the import, processing and visualization of my data.


My Discovery Summit poster shares more details about how I imported and prepared nearly four years of two different types of data collected with my activity monitor armband and its web- and app-based food logging software. I hope you will join me at the poster session on Wednesday, Sept. 17, to learn more about how I’ve used my own data to better understand the patterns in my weight loss and maintenance efforts.


If you stop by at my poster, you’ll also get a sneak peek at some of the new features coming in JMP 12 because I used many of them! If you are a member of the JMP User Community, you can see a PDF version of my poster on the JMP Discovery Summit 2014 community.  Those of you who know me are probably not surprised that this blog post was edited for length.  You can read a longer version of this post on my JMP User Community blog here. (Psst. It’s free and simple to become a member of the JMP User Community.)


Upcoming blog posts will share more about how I got my fitness and diet data into JMP and worked with JMP visualization expert and developer Xan Gregg to optimize visualizations that appear on my Discovery Summit poster.

Community Member

Michael Anderson wrote:

Thanks so much for sharing this Shannon - this was my favorite poster from the JMP Discovery conference! It is interesting the extent to which those of us involved with statistics and health & fitness do extensive self-tracking! I have specifically avoided the level of weight tracking you did for my own reasons ... but I also regularly run with 3 different types of trackers, track steps with a fitness band, monitor heart-rate, and so on. And we have JMP to help us track and make sense of it all!

Thanks again!

Shannon Conners wrote:

Thank you, Michael! I really appreciate your comment and I'm glad you enjoyed my poster. I think a lot of us who like collecting, visualizing and analyzing other kinds of data end up tracking fitness data with various kinds of monitors. I find I learn new things when exploring my own data in JMP, which I can then apply to work-related projects too. I was able to tackle several scripting projects at work after I completed this project that I could not have envisioned completing before. And of course, I learned some amazing tips and tricks for Graph Builder in the process!

Community Member

Martin wrote:


This is something I have a huge interest in. Would anyone be willing to share data as I think I have a way to extract much, much more information from this using an algorithm I created. I just lack data to put through it!

Shannon Conners wrote:

You can contact others interested in this topic on the forums at I know that some there and others online have made their QS data public. However, I think the best way to really understand the benefits and drawbacks to QS data collection and to test algorithms related to it is to collect your own data. Despite thinking I was fairly consistent with my own data collection, upon further examination, I noted inconsistencies and unexpected problems which might have been missed by someone else who was less familiar with how the data set was collected. Good luck!


I never tire of hearing about your weight loss journey!  I have known you for such a long time and you are truly an inspiration!   I love JMP too!!!   I have followed your data analysis project, using JMP,  with your weight loss from the beginning!    You are motivating ME again too!  Looks those guns girl!   Thanks for this story Shannon!